Moderation API

An AI-powered content moderation API that analyzes text for harmful content using multiple AI providers and two analysis modes.

Features

• Single mode — Analyzes a single text using Hugging Face's toxic-bert model
• Context mode — Analyzes a full conversation with context awareness using LLMs (Groq, Gemini, OpenAI)
• Multi-provider architecture — switch AI providers via environment variable
• Moderation history with filtering and pagination
• Statistics endpoint
• Rate limiting
• Swagger documentation

Requirements

• Node.js 18+
• PostgreSQL

Installation

1. Clone the repository

```bash git clone https://github.com/Aello1/moderation-api.git cd moderation-api ```

2. Install dependencies

```bash npm install ```

3. Create your .env file

```bash cp .env.example .env ```

4. Fill in your credentials in .env

5. Set up the database

```bash npx prisma db push ```

6. Start the server

```bash npm run dev ```

API Documentation

Swagger UI available at http://localhost:3000/docs

Usage

Single mode

```json POST /api/moderate { "mode": "single", "text": "I hate you" } ```

Context mode

```json POST /api/moderate { "mode": "context", "messages": [ { "username": "alice", "role": "user", "content": "hey" }, { "username": "bob", "role": "user", "content": "I will destroy you" } ] } ```

Get moderation logs

``` GET /api/logs?flagged=true&action=block&page=1&limit=10 ```

Get statistics

``` GET /api/stats ```

Configuration

Variable	Description	Default
DATABASE_URL	PostgreSQL connection string	—
HF_API_KEY	Hugging Face API key	—
GROQ_API_KEY	Groq API key	—
AI_PROVIDER	AI provider for context mode (groq)	groq
PORT	Server port	3000

Tech Stack

• TypeScript + Node.js
• Express
• PostgreSQL + Prisma
• Hugging Face — toxic-bert
• Groq — Llama 3.3 70B
• Swagger UI

Note: Gemini provider may not be available in all regions due to Google API restrictions.

License

MIT with Commons Clause — free for non-commercial use. Contact for commercial licensing.